The big graphics card review: from GTX 465 and HD 5830 to high-end

Our test configuration:

Motherboard:
- GIGABYTE GA-EX58-UD4P (BIOS: F13)
Processor:
- Intel Core i7 920 2,66 GHz @ 3,6 GHz (200×18)
Processor Cooling:
- Scythe Ninja 2 Rev B
Memory:
- Kingston HyperX T1 1600 MHz 9-9-9-24 2 × 2 GB 1,65 V
Hard drives:
- HITACHI 160 GB SATA2 (HDS721616PLA380)
- Samsung 200 GB SATA2 (SP2004C)
Graphic cards:
- GeForce GTX 465 (ASUS ENGTX465 1024 MB GDDR5)
- GeForce GTX 465 SLI (2 × ASUS ENGTX465 1024 MB GDDR5)
- GeForce GTX 470 (ASUS ENGTX470 1280 MB GDDR5)
- GeForce GTX 470 SLI (2 × ASUS ENGTX470 1280 MB GDDR5)
- GeForce GTX 480 (ASUS ENGTX480 1536 MB GDDR5)
- GeForce GTX 480 SLI (ASUS ENGTX480 1536 MB + GIGABYTE GTX480 1536 MB)
- ATI Radeon HD 5830 (ASUS EAH5830 DirectCU 1024 MB GDDR5)
- ATI Radeon HD 5830 CrossFireX (ASUS EAH5830 DCU + GIGABYTE HD 5830 UDV)
- ATI Radeon HD 5850 (ASUS EAH5850 1024 MB GDDR5)
- ATI Radeon HD 5850 CrossfireX (2 × ASUS EAH5850 1024 MB)
- ATI Radeon HD 5870 (ASUS EAH5870 1024 MB GDDR5)
- ATI Radeon HD 5870 CrossfireX (ASUS EAH5870 + GIGABYTE HD 5870)
Power Supply: Xigmatek NRP-HC1501 1500 W
Softwares:
- Windows 7 RTM 64 bit Ultimate HUN
- Intel INF 9.1.1.1019
- NVIDIA GeForce 257.21 WHQL x64
- ATI Catalyst 10.6 WHQL x64
- Realtek HD Audio 2.49 Driver
Display: ASUS 24T1 TV Monitor

Core i7 920 @ 3600 MHz + Kingston HyperX T1 1600 MHz CL8 + GIGABYTE GA-EX58-UD4P

Graphic drivers handle:

Before we let to lose test result, we need to clarify how we made it. We did not touch driver settings at synthetic measurements (3DMark Vantage, DirectCompute Benchmark, Unigine Heaven Benchmark, ComputeMark, MediaShow Ultra), everything ran at default mode. After synthetic measurements we turn on manually 16× anisothropic filtering at both drivers because we can not change that in a lot of application and we felt important for the testing.

Measurement of usage and warming:

We measured usage and warming at all participant. We checked appetite of cards with watt metre. Without loading we watched Windows’ calm state, with loading we used FurMark 1.8.2 built-in modul and ran 5 minutes at benchmark mode with the most agressive settings. At SLI and CrossFireX system we ran software at dual GPU mode, of course. We tried to form rough avarage at both state. We used Afterburner 1.6.0 for measure warming and fan’s rpm. We stared monitoring with calm state and we used FurMark for loading and we let the cards cool down at the finish.

Checking cards’ temperature datas, loadings and fans rotate numbers: